search

pydelhi / talks

Talks at Python Delhi User Group!
https://pydelhi.org/talks/
66 stars 51 forks source link

From Open Source Basics to Building with InstructLab #291

Closed 1610ayushi closed 2 weeks ago

1610ayushi commented 3 weeks ago

Title

From Open Source Basics to Building with InstructLab

Describe your Talk

In this session, we will dive into InstructLab, a new community-based approach to building truly open-source large language models (LLMs). We’ll begin by explaining the significance of open-source development in the AI landscape and how it empowers global collaboration. The focus will be on how InstructLab fosters a community-driven effort to develop LLMs that are accessible, transparent, and innovative. By simplifying the complex world of machine learning, InstructLab is enabling contributors to actively participate in the creation of powerful AI tools, pushing the boundaries of open-source AI.

Pre-requisites & reading material

Basic knowledge of LLM

Time required for the talk

15-20 mins

Link to slides/demos

https://docs.google.com/presentation/d/1d08YUbL8JlBPrDAryebkXDqYlBYPmCfuw2JYBxq1jNw/edit?usp=sharing

No response

About you

I’m Ayushi Tiwari, an Associate Software Engineer at Red Hat with over 1 year of experience in AI, machine learning, and open-source technologies. In addition to my engineering work, I am also a public speaker, having delivered talks at various colleges and events. I am passionate about sharing my knowledge in AI and ML, and I actively contribute to the open-source community, aiming to make cutting-edge technologies more accessible and understandable.

LInkedin - https://www.linkedin.com/in/ayushitiwari16/

Availability

24/08/2024

Any comments

No response

Animesh-Ghosh commented 3 weeks ago

Hi @1610ayushi

Could you also dive into one aspect of InstructLab's architecture, even if briefly, along with doing the community driven aspect?

I understand InstructLab is written in Python. If you do give an overview of its architecture, that would be more valuable and foster good discussions around your talk.

No need to drop the community aspect, but augmenting the talk with some technical depth will enhance audience takeaways.

1610ayushi commented 3 weeks ago

Okay sure. I will explain the architecture and a small demo that how to do synthetic data generation

Animesh-Ghosh commented 3 weeks ago

Please do share the slides for future speakers' reference @1610ayushi.

You may edit the initial comment itself and add the link there.

Animesh-Ghosh commented 2 weeks ago

Thanks for the talk @1610ayushi!

Looking forward to more such talks in the future.